Weitere ähnliche Inhalte Ähnlich wie Design of an IT Capstone Subject - Cloud Robotics (20) Mehr von ITIIIndustries (20) Kürzlich hochgeladen (20) Design of an IT Capstone Subject - Cloud Robotics1. IT in Industry, vol. 2, no. 3, 2014 Published online 27-Oct-2014
ISSN (Print): 2204-0595
Copyright © Authors 104 ISSN (Online): 2203-1731
Design of an IT Capstone Subject - Cloud Robotics
Kieren Lim, Jayantha Katupitiya
School of Mechanical & Manufacturing Engineering
The University of New South Wales
Sydney, Australia
j.katupitiya@unsw.edu.au
Ka Ching Chan, Mary Martin
Department of Computer Science & Computer Engineering
La Trobe University
Bendigo, Australia
ka.chan@latrobe.edu.au, m.martin@latrobe.edu.au
Abstract—This paper describes the curriculum of the three
year IT undergraduate program at La Trobe University, and the
faculty requirements in designing a capstone subject, followed by
the ACM’s recommended IT curriculum covering the five pillars
of the IT discipline. Cloud robotics, a broad multidisciplinary
research area, requiring expertise in all five pillars with
mechatronics, is an ideal candidate to offer capstone experiences
to IT students. Therefore, in this paper, we propose a long term
master project in developing a cloud robotics testbed, with many
capstone sub-projects spanning across the five IT pillars, to meet
the objectives of capstone experience. This paper also describes
the design and implementation of the testbed, and proposes
potential capstone projects for students with different interests.
Keywords—cloud robotics; networked robotics; Internet of
Things; capstone projects; IT education
I. INTRODUCTION
A. Capstone Subject
As defined by the FSTE (Faculty of Science, Technology,
and Engineering) at La Trobe University, a capstone is a
culminating subject within the final year of a discipline course.
A capstone subject provides opportunities for students to
synthesis and demonstrate discipline knowledge acquisition of
graduate capabilities, employment skills and areas of learning
acquired during their undergraduate degree [1]. At La Trobe
University, Information Technology is a three year
undergraduate degree with a strong focus on work integrated
learning [2]. Students are required to complete at least one
capstone subject in the final year. The design of a capstone
subject is the responsibility of the capstone subject coordinator;
guidelines for this are provided by the faculty. Most of the
design guidelines are assessment related. The design
requirements unrelated to assessments are given as follows.
a capstone subject must be designed to be authentic
within the discipline, often through the use of real-
world problem solving; and
integrate relevant subject elements especially where
these are delivered in different subjects.
In the next section we will discuss the IT curriculum and
explain how cloud robotics as final year projects meets the
above design requirements well.
B. IT Curriculum
The IT curriculum at La Trobe University is ACS
(Australian Computer Society) accredited, and follows the
latest 2008 curriculum guidelines for undergraduate degree
programs in IT, recommended by the ACM (Association for
Computing Machinery) [3]. The ACM report was an extensive
undertaking, with direct contribution by over thirty people and
widely reviewed by academics and practitioners. According to
this report, the academic discipline of IT can well be
characterized as the most integrative of the computing
disciplines. The depth of IT lies in its breadth: an IT graduate
needs to be broad enough to recognize any computing need and
know something about possible solutions. The IT graduate
would be the one to select, create or assist to create, apply,
integrate, and administer the solution within the application
context [3]. This report also suggested the five pillars of the IT
discipline:
1. programming
2. networking
3. human-computer interaction
4. databases
5. web systems
The five pillars should be built on a foundation of
knowledge of the fundamentals of IT. Overarching the entire
foundation and pillars are information assurance and security,
and professionalism.
Fig. 1 depicts the IT curriculum at La Trobe University. It
consists of four fundamental IT subjects, two or more subjects
in each of the five pillars, one subject in Information and IT
security, one subject in Professional Development, electives,
industry based learning, minor and major projects. The final
year one semester project has been chosen as the capstone
subject for the degree.
Cloud robotics, coined by James Kuffner of Google in 2010
[4], aims to remove conventional limitations and introduce
greater capabilities through applying the benefits of cloud
technology to robotics and automation. Any application of
cloud robotics requires in-depth understanding and integration
of all five pillars of the IT discipline with mechatronics. IT’s
broad nature makes cloud robotics an ideal candidate to
become an IT capstone. This paper proposes to develop a cloud
robotic testbed as a long term master project, with many sub-
2. IT in Industry, vol. 2, no. 3, 2014 Published online 27-Oct-2014
ISSN (Print): 2204-0595
Copyright © Authors 105 ISSN (Online): 2203-1731
Fig. 1. Curriculum of La Trobe University’s Bachelor of Information Technology.
projects offered as capstone projects to final year IT students.
Through the capstone experiences, students will develop
expertise in their areas of interests. At the same time, working
on a project as part of a larger project with a common theme
will help students understand the bigger pictures, as well as the
connections of the pillars, and current and emerging IT
technologies.
The cloud testbed has been constructed for both teaching
and research purposes. Its requirements:
Low cost: Components should be open-source, off-the-
shelf, and inexpensive.
Flexible: Various types of hardware and software
should be easily integrated into the testbed.
Adaptable: The testbed should be able to accommodate
various networking methods and protocols.
Accessible: The testbed should be comprised of
relatively common components so as not to complicate
any results produced.
Open source software, off-the-shelf electronic and
mechanical components, DIY robotic kits, open source system-
on-a-chip (SoC), low cost wireless communications
components such as Wi-Fi and ZigBee, and low cost mobile
devices, have gradually become commodities in the
marketplace and readily available through online shops around
the world. This continuous downward trend on pricing has
made the development of such a cloud robotics testbed
affordable and attractive from an educator’s perspective.
II. CLOUD ROBOTICS
Traditional robotics was characterised by computerised
systems of sensors and actuators performing various functions,
facilitated by onboard computation, programming and data
storage. These robots usually have limited capabilities, operate
as standalone machines, and lack the ability to network with
other devices. As such, a solution was proposed in the form of
networked robots. The IEEE Robotics and Automation Society
defined "networked robot" as a robotic device connected to a
communications network such as the Internet or LAN [5]. With
this extended connectivity, the functionality of robots extended
to tasks no single robot could accomplish.
Networked robots have significant advantages as any single
robot has access to the computational power and data storage
of the entire network of robots resulting in improved task
efficiency. A major advantage is that sensor data of every
device in the network can be shared [6]. Networked robots are
likened to animals working in groups, enabling behaviours far
more complex and intelligent than is possible by any
individual.
3. IT in Industry, vol. 2, no. 3, 2014 Published online 27-Oct-2014
ISSN (Print): 2204-0595
Copyright © Authors 106 ISSN (Online): 2203-1731
The modern trend of cloud computing adds another
dimension to networked robotics. Cloud computing is a term
used to refer to the use of modern networking resources to
develop dynamic computing solutions by reducing limitations
in areas such as data storage and computational power. This is
accomplished through interaction with a significant number of
computers and servers, often represented by virtual hardware,
which introduces benefits such as reduced costs in electricity
and network bandwidth [7]. Enabling robots access to these
cloud computing resources led to the development of cloud
robotics. With cloud computing providing computing resources
on demand, robots would have the option of offloading
computationally intensive tasks to these resources.
Robots with access to the cloud no longer require expensive
processors for heavy computations as they can be offloaded.
Operations like image processing, voice recognition, 3D
mapping or other uncertainty-plagued tasks can use parallel
processing in the cloud and the results returned to the robot. In
addition, tasks such as motion planning [8] and control [9] can
also be performed in the cloud. All these strategies result in
lower power consumption and lighter hardware requirements.
Goldberg and Kehoe [10] identify further potentials of
cloud robotics:
Access to global libraries of images, maps and object
data
Robot sharing of outcomes, trajectories and dynamic
control policies
On-demand human guidance
Going beyond relational databases, through the cloud a
robot has access to ever-growing datasets of images, maps and
object data, also known as “Big data” [11]. With access to such
extensive data, a variety of applications are possible. One
example is the use of cloud datasets to facilitate robot grasping
[12], a task which requires extensive data on objects, their
features and how to manipulate them. Since a robot could be
asked to grasp any object, access to the cloud gives the robot
access to constantly updated object and grasp data. This
operation is further enabled by the parallel processing
capability of cloud computing to evaluate algorithms and
determine optimal grasps. In particular, parallel processing is
advantageous in situations with shape uncertainty.
As robots or other devices and systems perform operations,
their results become accessible to other devices through the
cloud. This opens the possibility of learning new functions and
capabilities based on successfully performed operations by
other robots. Extending the grasping example, a successful
grasp on a particular object performed by another robot can be
downloaded and performed in the same context removing
uncertainty and the need to employ intensive computations.
Path planning is another suitable candidate as robots can use
paths previously determined as successful. This concept of
shared capabilities and robot learning provides a significant
advancement in robot development.
Goldberg and Kehoe [10] suggest a crowdsourcing
situation where, in particular situations, robots seek the
assistance of humans. This enables human intervention on
tasks that may require human intuition or complex decision-
making. This is particularly relevant to dynamic environments
in which robots often encounter difficulties. Questions arise
about which events require human guidance; however, it is
evident that this ability is a desirable consequence of cloud
robotics.
Furthermore, crowdsourcing enhances human-robot
interaction, an area of increasing relevance as interest in
service robots and the Internet of Things (IoT) [13] continues
to grow. Rusu et al. [14] developed a service robot system,
based on ubiquitous computing, designed to operate in a
kitchen and perform common human tasks. This system is
characterised by distributed and embedded computing devices
throughout the environment, such as Radio Frequency
Identification (RFID) tags, which allow the robot to identify
objects.
The desirability of affordable robotic devices with
consistently supported software, in addition to the current trend
of enabling most electronic consumer products with network
connectivity, will result in a portion of the IoT consisting of
robots [15]. Coordinated interaction between IoT and cloud
robots could include utilization of data from environmental
sensors to achieve the localization of a robot with no onboard
sensors [16]. IoT architecture and Machine to Machine (M2M)
communication will enable cloud robotics to perform many
tasks with minimal human interaction [17]. Consider a scenario
in which a cloud robot identifies a faulty component, which is
then automatically printed by a nearby 3D printer and finally
installed by the human user in a similar fashion to Swedish
furniture.
It has become apparent that a number of current and
emerging technologies, such as big data, IoT, M2M, wireless
sensor networks, mesh networking, mobile computing, and
cloud computing, etc are converging at a fast pace, enabling.
innovative applications in multiple industries, such as health,
manufacturing, farming and agriculture. It is also not difficult
to imagine one day a human-robot social network making new
things and doing fun things together.
Two state-of-the-art cloud robotics developments worth
mentioning are Willow Garage and Google cooperating on
porting Willow Garage’s Robot Operating System (ROS) to
Android devices enabling the development of robots that could
interact with the cloud [18] and RoboEarth.org creating a
database of reusable skills thus facilitating one of the key
concepts of cloud robotics [19, 20].
III. CLOUD ROBOTIC TESTBED DESIGN
Cloud robotic systems are complex entities requiring
multiple elements and layers for operation. The client-server
system architecture, illustrated in Fig. 2, indicates the cloud is
the central component through which communication and
access requests are made. Furthermore, it must be capable of
extensive computations and hold data to be shared amongst
robots and users. Design and implementation of a highly-
available cloud platform, using open source software such as
OpenStack [21] or Proxmox [22], to support our cloud robotics
research are ideal capstone project topics.
4. IT in Industry, vol. 2, no. 3, 2014 Published online 27-Oct-2014
ISSN (Print): 2204-0595
Copyright © Authors 107 ISSN (Online): 2203-1731
Fig. 2. Client-server cloud robot architecture.
An open source Apache web server satisfies the
requirement of accessibility, providing avenues for
communication such as an API or web service. Furthermore, it
is accessible by any device with Internet capabilities, extending
access beyond PCs or robots to other embedded devices such
as smartphones. Design and building additional web services
and applications for increased cloud functionality are also ideal
capstone projects. For the testbed to be operated on common
platforms and simplified for users, MySQL [23] and Structured
Query Language (SQL) were selected.
To allow multiple robotic platforms to perform within the
testbed, Chen and Bai [24], Johnson et al. [25], and Kato et al.
[26] demonstrated intermediate platforms within the cloud to
track and manage all the robots in operation. The robot
manager performs this function. It provides task management
capabilities if necessary, such as coordinating multiple robots
for a single service, however, in simple operation it acts as the
interface between the cloud and robotic devices.
In this model, the cloud is a server to which robot clients
must connect. A resource-based architecture was ideal to abide
by Service-Oriented Architecture (SOA) design. This meant
robot clients must be abstracted into resources accessible by the
server. However, the question arose as to the level of
abstraction necessary for this project. For example, a robot
could simply be presented as a device with the capabilities of
proximity detection and motion, or it could be presented as
multiple sensors and motors, each with its own functions.
Alternatively, if grouping of sensors is desired then it may
occur at different levels such as grouping all infrared sensors
together and all ultrasonic sensors together or grouping sensors
according to their location on a device. It was decided that, in
order to enable this architecture to operate with differing
hardware platforms, resource abstraction should occur on the
robot before notifying the server. Presenting each robot as a
single resource, while significantly reducing complexity in the
system, would limit functionality and flexibility as users could
not take advantage of individual sensors or robot capabilities.
As a result, in this project, each individual robotic component
is considered a resource.
In a similar manner to resource abstraction, control
operations may differ on different devices due to various
hardware configurations. It was decided that control would be
performed on the robot itself.
The front-end of the system enables users to access the
testbed. For experimentation, user interaction is necessary as
opposed to simply viewing the system. A user-facing API was
desirable as provides an interface for researchers to build
applications upon the resources within the testbed. This API
was to be built upon a web service framework.
A. Robot Setup
Initially there are two robots, of varying configuration and
different levels of capabilities, in the testbed, but these will
grow. The first robot built consists of multiple sensors to
identify its surrounding environment. These components are
placed in a configuration upon the chassis, consisting of motors
for robot motion. The array of sensors selected was to
demonstrate both operation of a multi-sensor robot and to show
integration of multiple types of sensors into this system. As
indicated in Fig. 3, three types of sensors were chosen:
1. Inertial Measurement Unit (IMU): IMUs are available
with various components such as an accelerometer,
gyroscope and magnetometer. This is useful for
determining the location and orientation of a robotic
platform.
2. Infrared (IR): These proximity sensors emit a beam of
light which reflects off any object in its line of sight to
a receiver. IR sensors were placed at the front-left and
front-right of the robot.
3. Ultrasonic: Another type of proximity sensors that emit
an ultrasonic pulse which bounces off objects and
returns to a receiver. An ultrasonic sensor was placed
centrally at the front of the robot.
The second robot has no proximity sensors but, other than
motors for motion, only has an IMU to identify its location and
orientation. The inability to identify its environment through
interpretation of sensor data causes it to be completely reliant
on information received from the cloud. In this system, robots
of this kind may function, provided there is another source
relaying data of the surroundings, such as another robot with
proximity sensors or an online map.
Both robots followed the same basic design. The first step
was building the chassis followed by installing the Raspbian
operating system [27] on the Raspberry Pi [28]. Subsequently,
the GertDuino expansion board [29] was initialised and
5. IT in Industry, vol. 2, no. 3, 2014 Published online 27-Oct-2014
ISSN (Print): 2204-0595
Copyright © Authors 108 ISSN (Online): 2203-1731
Fig. 3. Robot configuration
interfaced with the Raspberry Pi through serial communication.
Then the Ardumoto motor shield [30] was connected and
control of the motors was established. Finally, the sensors were
configured and calibrated to provide accurate measurements.
As IT capstone projects, mechanical design of robots is not
the main focus. Project students are given the flexibility to
design and buid their own robots, or to hack and modify those
remote control toy cars, trucks, or tanks. The testbed will
consist of an army of different types of robots, which may have
similar or different designs and functionalities, working
together to achieve certain purposes. In addition to the two
basic robots, as shown in Fig. 4, one student has built a two-
wheeled self-balancing robot; and another has hacked a low
cost remote control toy racing car. These robots are very
different in design and require different control algorithms.
However, similar to the two basic robots, they are also
controlled by a Raspberry Pi and Arduino. We have found that
hacking low cost remote control cars is an attractive option for
students. There are many toy cars with creative and interesting
designs available in toy shops or supermarkets. Students enjoy
the work involved in hacking and controlling them with their
own systems.
B. I/O Expansion
The GertDuino board is an expansion board to interface
between the Raspberry Pi and Arduino Uno compatible shields.
It contains two microcontrollers: the Atmega328, the same
microcontroller as the Arduino Uno, and an Atmega48.
This expansion board increases the current capabilities of
the Raspberry Pi as it provides access to an Analog-to-Digital
Converter (ADC), for 6 analog inputs, and 14 digital outputs,
of which six can be used for pulse width modulation (PWM)
output. Additionally, the GertDuino enables interfacing with
established Arduino hardware and software and the extensive
Arduino community.
Using a custom version of AVRDUDE provided by Gordon
Henderson, the Atmega328 microcontroller was configured to
Fig. 4. Robot 1 and robot 2.
run at 16MHz to enable compatibility with the Arduino Uno
[30]. With initial setup complete, it was possible to program
the Atmega328 through the Arduino IDE on the Raspberry Pi.
However, the program would run independently of the
Raspberry Pi.
To take advantage of the capabilities of the Raspberry Pi
and any connected Arduino shields, communication is required
between the devices. Through the GertDuino, a serial
connection between the Raspberry Pi and Atmega328 was
established [28, 32, 33]. The Atmega328 was programmed
using the Arduino Serial library and the pySerial package for
Python 2.7. With the motors and sensors configured, it was
possible for the Raspberry Pi to control the motors and receive
sensor data.
C. Networking and Communications
To initiate wireless communication, Universal Serial Bus
(USB) Wi-Fi dongles were used for both robots and individual
static IP addresses configured for each robot. This provided
LAN and Internet access to the robots, and allowed for a virtual
network computing (VNC) server to be set up on each
Raspberry Pi. The VNC server allows a PC wireless access to
the robots and removes the need for peripherals, such as a
monitor or keyboard. Changes in software and control can be
performed remotely, significantly easing testing and
operations. TightVNC software was used to set up both the
VNC servers on the robots and the client on the PC [34].
With Wi-Fi setup on each robot, the testbed allows students
interested in networking to experiment with MANET (Mobile
Ad Hoc Network) and VANET (Vehicular Ad Hoc Network),
and wireless mesh network routing protocols such as OLSR,
Batman, and Babel. Other wireless communications
technologies, such as IPv6 over Low power Wireless Personal
Area Networks (6LowPan) and the IEEE 802.15 ZigBee
networks, also have roles to play in Cloud Robotics. Other
potential capstone projects include design, implement, and
evaluate Voice over IP (VoIP) and video streaming
technologies over different network environments. An example
of such applications can be found in [35]. Students will
develop their own communication systems using the same core
VoIP technology, but with their own creative design of
graphical user interfaces for remote control using a web
browser and/or an Android tablet or phone.
6. IT in Industry, vol. 2, no. 3, 2014 Published online 27-Oct-2014
ISSN (Print): 2204-0595
Copyright © Authors 109 ISSN (Online): 2203-1731
D. Cloud Infrastructure
To support the long term development of the testbed, the
open source cloud platform OpenStack, was specifically
designed as part of the whole project. The whole platform has
been setup in La Trobe University’s Internetworking lab [36].
The cloud computing platform leverages the power of
virtualization and redundancy to provide highly available
computing infrastructure to multiple robots, and host all the
required backend systems and services. The platform also
provides varieties of robots an ideal, centralized data storage
platform for communication and cooperative control.
The infrastructure includes a number of Cisco 2801 routers,
Cisco 2960 switches, Cisco Aironet access points, Cisco
Catalyst 3750 series wireless LAN controllers, purpose built
Intel i-7 servers with a minimum of 24GB of RAMs, and
QNAP TS-410 1U Networked Storage Servers (NAS).
Networking Students are involved in aspects of the design,
setup, and evaluation of high available network and cloud
infrastructures to support the testbed.
E. Programming and Control
Within the proposed architecture, there are two key
interfaces where connectivity and communication are required:
the cloud-user interface and robot-cloud interface. A client-
server model was selected for interaction throughout the
architecture with the cloud acting as centralised servers and the
robots being clients. At the other interface, the users would act
as clients to the cloud server. This forms the basis of
communication as, for this testbed operation, a connection with
the cloud is essential for each device and is the most suitable
with a limited number of devices. For further development,
with a greater number of robots, a peer-to-peer topology could
be established with modest effort. In a similar scenario, a
publish-subscribe communication model could be presented.
The HTTP protocol is used for data and message transfer in
API and web application development, largely due to its
relative lack of restrictions [37]. In API development, there are
two widely used frameworks for using HTTP, REST and
SOAP [38]. REST is able to communicate through multiple
data formats, while SOAP uses a standard format in XML. In a
comparison of JSON and XML, JSON was found to be ideal
for this type of project as it is populated with resource limited
devices [39]. Hence JSON was used for data transfer
throughout this project; and consequently the REST framework
was implemented for an API between the cloud and users. In
addition, it allowed a web service to be stateless meaning client
requests had to be explicit and URIs had to provide a
comprehensive method of resource identification. Using
similar technology, the Robot-Cloud interface was designed as
a web service on the server side to provide connections to the
robot clients. For programming students abundant capstone
opportunities are available in areas such as API development,
mobile app development, web services and cloud based control
algorithms.
IV. CONCLUSION
This paper described the curriculum of the three year IT
undergraduate program at La Trobe University, and the
faculty’s requirements in designing a capstone subject. As the
broadest and most integrative of the computing disciplines,
ACM recommended the five pillars of the IT discipline. We
have proposed to develop a cloud robotics testbed, as a master
project, with many capstone sub-projects spanning across the
five pillars. We have presented the initial design and
implementation of the testbed, which will continue to offer
abundant opportunities to students to gain valuable capstone
experiences.
ACKNOWLEDGEMENT
The authors would like to thank La Trobe University for the
financial support through a special research grant scheme to
aid regional research. This paper is an extended version of a
paper previously presented at the 9th
International Conference
on Information Technology and Applications (ICITA), in
Sydney, 1-4 July 2014.
REFERENCES
[1] E. Johnson, “Capstone subjects in FSTE,” unpublished, Faculty of
Science, Technology and Engineering, La Trobe University, 2013.
[2] L. Staehr, M. Martin, and K. Chan, “A multi-pronged approach to work
integrated learning for IT students,” J. Information Technology
Education: Innovations in Practice, vol. 13, pp. 1-11, 2014.
[3] B. M. Lunt, J. J. Ekstrom, S. Gorka, G. Hislop, R. Kamali, E. Lawson,
R. LeBlanc, J. Miller, and H. Reichgelt, “Information Technology, 2008,
Curriculum guidelines for undergraduate degree programs in
Information Technology,” Association for Computing Machinery
(ACM) and IEEE Computer Society, 2008. [Online]. Available:
http://www.acm.org/education/curricula/IT2008%20Curriculum.pdf
[4] J. Kuffner, “Cloud-Enabled Robots,” in IEEE-RAS Int. Conf. Humanoid
Robots, Nashville, TN, 2010.
[5] IEEE RAS Technical Committee. (2012, September). IEEE Society of
Robotics and Automation's Technical Committee on: Networked Robots
[Online]. Available: http://www-users.cs.umn.edu/~isler/tc/
[6] V. Kumar, G. Bekey, and A. Sanderson, “Networked Robots,” in
Assessment of International Research and Development in Robotics,
NASA, 2006, pp. 73-80.
[7] M. Armbrust, A. Fox, R. Griffith, A. D. Joseph, R. Katz, A. Konwinski,
et al., “A view of cloud computing,” Communications of the ACM, vol.
53, pp. 50-58, 2010.
[8] Y. Mei, Y.-H. Lu, Y. C. Hu, and C. G. Lee, “Energy-efficient motion
planning for mobile robots,” in Proc. IEEE Int. Conf. Robotics and
Automation, ICRA'04. 2004, pp. 4344-4349.
[9] A. Barili, M. Ceresa, and C. Parisi, “Energy-saving motion control for
an autonomous mobile robot,” in Proc. IEEE Int. Symp. Industrial
Electronics, ISIE'95. 1995, pp. 674-676.
[10] K. Goldberg and B. Kehoe, “Cloud robotics and automation: A survey
of related work,” EECS Department, University of California, Berkeley,
Tech. Rep. UCB/EECS-2013-5, 2013.
[11] J. Manyika, M. Chui, B. Brown, J. Bughin, R. Dobbs, C. Roxburgh,
Angela Hung Byers. (2011, May). Big data: The next frontier for
innovation, competition, and productivity [Online]. Available:
http://www.mckinsey.com/insights/business_technology/big_data_the_n
ext_frontier_for_innovation
[12] B. Kehoe, A. Matsukawa, S. Candido, J. Kuffner, and K. Goldberg,
“Cloud-based robot grasping with the google object recognition engine,”
in IEEE Int. Conf. on Robotics and Automation, 2013.
[13] L. Atzori, A. Iera, and G. Morabito, “The Internet of Things: A survey,”
Computer Networks, vol. 54, pp. 2787-2805, 2010.
[14] R. B. Rusu, B. Gerkey, and M. Beetz, “Robots in the kitchen: Exploiting
ubiquitous sensing and actuation,” Robotics and Autonomous Systems,
vol. 56, pp. 844-856, 2008.
[15] G. Hu, W. P. Tay, and Y. Wen, “Cloud robotics: architecture, challenges
and applications,” Network, IEEE, vol. 26, pp. 21-28, 2012.
7. IT in Industry, vol. 2, no. 3, 2014 Published online 27-Oct-2014
ISSN (Print): 2204-0595
Copyright © Authors 110 ISSN (Online): 2203-1731
[16] L. Wang, M. Liu, M. Q. H. Meng, and R. Siegwart, “Towards real-time
multi-sensor information retrieval in Cloud Robotic System,” in 2012
IEEE Conf. Multisensor Fusion and Integration for Intelligent System,
MFI. 2012, pp. 21-26.
[17] L. Tan and N. Wang, “Future internet: The internet of things” in 2010
3rd
Int. Conf. Advanced Computer Theory and Engineering, ICACTE.
2010, pp. V5-376.
[18] Willow Garage. (2013, December) PR2. [Online]. Available:
https://willowgarage.com/pages/robots/pr2-overview
[19] M. Waibel, M. Beetz, J. Civera, R. D'Andrea, J. Elfring, D. Galvez-
Lopez, et al., “Roboearth,” Robotics & Automation Magazine, IEEE,
vol. 18, pp. 69-82, 2011.
[20] The RoboEarth Consortium. (2013, September). What is RoboEarth?
[Online]. Available: http://roboearth.org
[21] OpenStack. (2014, June). OpenStack Cloud Software. [Online].
Available: http://www.openstack.org
[22] Proxmox Server Solutions. (2014, June). Proxmox. [Online]. Available:
http://www.proxmox.com
[23] Oracle Corporation. (2014, June). Mysql The world’s most popular open
source database. [Online]. Available: http://www.mysql.com
[24] Y. Chen and X. Bai, “On robotics applications in service-oriented
architecture,” in 28th Int. Conf. Distributed Computing Systems
Workshops, ICDCS'08. 2008, pp. 551-556.
[25] D. Johnson, T. Stack, R. Fish, D. M. Flickinger, L. Stoller, R. Ricci, and
J. Lepreau, “Mobile emulab: A robotic wireless and sensor network
testbed,” in IEEE INFOCOM, 2006.
[26] Y. Kato, T. Izui, Y. Tsuchiya, M. Narita, M. Ueki, Y. Murakawa, et al.,
“RSi-cloud for integrating Robot Services with internet services,” in
IECON 2011-37th Annu. Conf. IEEE Industrial Electronics Society,
2011, pp. 2158-2163.
[27] Raspbian. (2014, April). Welcome to Raspbian. [Online]. Available:
http://www.raspbian.org/
[28] Raspberry Pi Foundation. (2014, March). About Us. [Online]. Available:
http://www.raspberrypi.org/about
[29] G. van Loo. (2014, March). GertDuino User Manual. [Online].
Available:
http://www.element14.com/community/servlet/JiveServlet/downloadBo
dy/64534-102-2-287165/User%20manual%20Gerduino%205.6.pdf
[30] SparkFun Electronics. (2014, March). Ardumoto – Motor Driver Shield.
[Online]. Available: https://www.sparkfun.com/products/9815
[31] G. Henderson. (2014, April). ATmega Setup. [Online]. Available:
https://projects.drogon.net/raspberry-pi/gertduino/atmega-setup/
[32] Friends of the Unicorn. (2014, April). Raspberry Pi-GertDuino Serial.
[Online]. Available: http://friendsoftheunicorn.net/content/raspberry-pi-
gertduino-serial
[33] CE Linux Forum. (2014, April). RPi Serial Connection. [Online].
Available: http://elinux.org/RPi_Serial_Connection
[34] TightVNC. (April 2014). TightVNC Software. [Online]. Available:
http://www.tightvnc.com/
[35] K. C. Chan, J. Katupitiya, J. W. Hanrahan, C. J. Jackson, D. Rose, “An
emergency communication system for an agricultural autonomous
vehicle,” in Proc. Int. Conf. Intelligent Agriculture, IPCBEE, vol. 63,
2014, pp. 76-82.
[36] K. C. Chan and M. Martin, “An integrated virtual and physical network
infrastructure for a networking laboratory,” in Proc. 7th
Int. Conf. on
Computer Science and Education, ICCSE. Melbourne, Australia, 2012,
pp. 1433-1436.
[37] K. W. Ross and J. F. Kurose, Computer networking: a top-down
approach featuring the Internet, 6th
ed. England: Pearson Education
Limited, 2013.
[38] C. Pautasso, O. Zimmermann, and F. Leymann, “Restful web services
vs. big'web services: making the right architectural decision,” in Proc.
17th Int. Conf. World Wide Web, 2008, pp. 805-814.
[39] N. Nurseitov, M. Paulson, R. Reynolds, and C. Izurieta, “Comparison of
JSON and XML Data Interchange Formats: A Case Study,” Caine, vol.
9, pp. 157-162, 2009.